Introduction
Stocks form a major part of the liquid economy of the world. The movement of stocks governs the country’s economy to a great extent. Capitalizing on this asset can result in major returns for a person/organization. The stock market
is also a very volatile system, with rate fluctuation from few rupees to thousands. Predicting such a system can be challenging and rewarding for an individual/organization who can capitalize on the trends seen in the stock market.
Traditional methods of Stock Market prediction involve a regression model on historical quotes (stock prices) which form a time series. Models like MARS, ARMA, ARIMA were frequently used. A common conclusion of all these researches was that the time series obtained from stock market prices follows the random walk theory. To move from the random walk theory to semi-strong Efficient Market Hypothesis (EMH) we need to take into account addition factors on top of the basic time series model. [1]
It has been shown that the financial market is “informationally efficient” [2] - stock prices reflect all known information, and the price movement is in response to news or events. As web information grows, recent work has applied Natural Language Processing (NLP) techniques to explore financial news for predicting market volatility.
Correlating financial news to the current stock market prices has been an successful additional factor to the traditional time series model. The basic approach to incorporating financial news in stock market prediction is classifying the direction of the stock market on the basis of a score (sentiment analysis) given to the news articles whose value deprecates over time. Classification algorithms such as genetic algorithms, Naive Bayes, and SVM are popular approaches.[3] A newer approach to the problem has been, using financial news articles to predict discrete values of the stock market prices. This is done by representing the news articles as a bag of words/noun phrases/named entities/events. Machine learning algorithms such as SVM, ANN and Deep Learning are popular approaches to this problem.[4] Researchers are also moving towards using social media in stock market prediction to understand the impact of a news on the audience which directly affect the stock market.
is also a very volatile system, with rate fluctuation from few rupees to thousands. Predicting such a system can be challenging and rewarding for an individual/organization who can capitalize on the trends seen in the stock market.
Traditional methods of Stock Market prediction involve a regression model on historical quotes (stock prices) which form a time series. Models like MARS, ARMA, ARIMA were frequently used. A common conclusion of all these researches was that the time series obtained from stock market prices follows the random walk theory. To move from the random walk theory to semi-strong Efficient Market Hypothesis (EMH) we need to take into account addition factors on top of the basic time series model. [1]
It has been shown that the financial market is “informationally efficient” [2] - stock prices reflect all known information, and the price movement is in response to news or events. As web information grows, recent work has applied Natural Language Processing (NLP) techniques to explore financial news for predicting market volatility.
Research Approaches
This blog discusses some of the NLP techniques that have been applied by researchers to improve the stock market predictions.Correlating financial news to the current stock market prices has been an successful additional factor to the traditional time series model. The basic approach to incorporating financial news in stock market prediction is classifying the direction of the stock market on the basis of a score (sentiment analysis) given to the news articles whose value deprecates over time. Classification algorithms such as genetic algorithms, Naive Bayes, and SVM are popular approaches.[3] A newer approach to the problem has been, using financial news articles to predict discrete values of the stock market prices. This is done by representing the news articles as a bag of words/noun phrases/named entities/events. Machine learning algorithms such as SVM, ANN and Deep Learning are popular approaches to this problem.[4] Researchers are also moving towards using social media in stock market prediction to understand the impact of a news on the audience which directly affect the stock market.
References
[1] http://www.johnwittenauer.net/a-simple-time-series-analysis-of-the-sp-500-index/
[2] Eugene F Fama. The behavior of stock-market prices. The journal of Business
[3] https://www.econ.berkeley.edu/sites/default/files/Selene%20Yue%20Xu.pdf
[4] http://www.ijcai.org/Proceedings/15/Papers/329.pdf
[2] Eugene F Fama. The behavior of stock-market prices. The journal of Business
[3] https://www.econ.berkeley.edu/sites/default/files/Selene%20Yue%20Xu.pdf
[4] http://www.ijcai.org/Proceedings/15/Papers/329.pdf
Comments
Post a Comment